Tumour necrosis as assessed with ¹⁸F-FDG PET is a potential prognostic marker in diffuse large B cell lymphoma independent of MYC rearrangements

For this retrospective single-centre study, consecutive patients with newly diagnosed, histologically confirmed DLBCL between 2008 and 2015 were identified in the electronic healthcare database of the University Medical Center Groningen (UMCG), a reference centre for aggressive B cell lymphomas. Cases of primary cutaneous DLBCL, primary central nervous system lymphoma, primary mediastinal B cell lymphoma and immunodeficiency-associated lymphomas were excluded. The selection of cases for this study is summarised in Fig. 1. Patients were stratified according to the National Comprehensive Cancer Network international prognostic index (NCCN-IPI) [22]. End of treatment response was assessed by ¹⁸F-FDG PET/CT scan. Tumour responses were classified according to Lugano criteria [12]. Follow-up was registered until early October 2017. According to Dutch regulations, no medical ethical committee approval was required for this retrospective, non-interventional study. A waiver was obtained from the medical ethics committee of the UMCG on November 13, 2018. The study utilised rest material from patients, the use of which is regulated under the code for good clinical practice in the Netherlands and does not require informed consent in accordance with Dutch regulations.

Pathology review was done using the 2008 WHO classification of haematopoietic and lymphoid tissues (AD) [23]. Histological scoring for necrosis (necrosis^Hist) was done by microscopic assessment of haematoxylin and eosin–stained slides. Only microscopic areas with definite histopathological signs of necrosis (i.e. karyolysis) were scored as positive for necrosis^Hist.

For evaluation of a MYC rearrangement, formalin-fixed paraffin-embedded tissue blocks of primary tumour samples were used. Interphase fluorescence in situ hybridisation (FISH) was performed on 4-μm-thick whole tissue sections, using Vysis break apart probes (Abbot Technologies) and standard FISH protocols as previously described [24]. Researchers performing MYC FISH analyses were blinded for results from visual scoring, microscopic assessment of necrosis (necrosis^Hist) and clinical outcome.

All ¹⁸F-FDG PET scans were performed prior to therapy. Patients were allowed to continue all medication and fasted for at least 6 h before whole-body (from the skull vertex to mid-thigh level) three-dimensional PET images were acquired. This was done 60 min after intravenous administration of a standard dose of 3 MBq/kg (0.081 mCi/kg) bodyweight ¹⁸F-FDG on a Biograph mCT (Siemens Healthineers), according to the European Association of Nuclear Medicine (EANM) procedure guidelines for tumour imaging with FDG PET/CT (version 2.0) [25]. Acquisition was performed in seven bed positions of 2-min emission scans for patients 60–90 kg. Patients with body weight less than 60 kg and more than 90 kg body weight were scanned with 1 min and 3 min per bed position, respectively. Low-dose transmission CT was used for attenuation correction. Low-dose CT and ¹⁸F-FDG PET scans were automatically fused by the use of three-dimensional fusion software (Siemens Healthineers) with manual fine adjustments. Raw data were reconstructed through ultra-high definition (Siemens Healthineers).

Diagnostic CTs were acquired via integrated ¹⁸F-FDG PET/CT scans according to the European Association of Nuclear Medicine (EANM) procedure guidelines for tumour imaging with FDG PET/CT (version 2.0) [25]. Bulky disease was defined as any nodal lymphoma lesion > 10 cm in coronal, axial or sagittal planes.

All ¹⁸F-FDG PET scans were visually assessed for the presence of tumour necrosis (necrosis^PET) by an experienced reader (TCK), who was blinded to clinical, laboratory, biopsy and follow-up findings, as previously described [15]. Areas within any nodal or extranodal ¹⁸F-FDG PET–avid lymphomatous lesions that showed no ¹⁸F-FDG uptake were registered as having necrosis^PET (Fig. 2); no specific visual scale was used. Semiquantitative analysis was performed using an in-house tool for quantitative ¹⁸F-FDG PET/CT analysis, as previously described [26‐28]. This programme automatically preselects lesions using a SUV_max threshold of 4 and a metabolic volume threshold of 2.5 ml. Unwanted preselected FDG-avid regions, such as the bladder and brain, are removed by user interaction. Finally, remaining FDG-avid segmentations are processed using a background-corrected 50% of SUV peak region growing method, as described by Frings et al [26], to obtain the final tumour segmentations. In case obvious lymphoma lesions were not selected (n = 3), they were manually added after automatic tumour segmentation. From the final segmentation, the metabolic active tumour volume (MATV, in ml), total lesion glycolysis (TLG = MATV × SUV_mean) and SUVs are derived for each lesion independently as well as summed over all lesions. Lesion selection and semiquantitative analysis was performed by MH under direct supervision of an experienced nuclear medicine physician (WN) and a nuclear physicist (RB). SUV_max was defined as the highest SUV per voxel within one lymphomatous lesion. In this paper, SUV_max is reported as the mean of SUV_max across all lesions of an individual patient. SUV_max single highest was defined as the highest SUV_max of all lesions within an individual patient.

Comparison between continuous, non-normally distributed variables was estimated by Wilcoxon rank-sum test. Differences between two nominal variables were evaluated using Pearson’s chi-square or Fisher’s exact test (for expected groups sizes ≤ 5). For exploratory survival analysis, the primary endpoints were overall survival (OS), progression-free survival (PFS) and disease-specific survival (DSS). OS was defined as the time from diagnosis until death (from any cause). PFS was defined as the time from diagnosis until death or relapse or progression [12]. DSS was defined as the time from diagnosis until death from DLBCL. Surviving patients were censored at the last date of follow-up. Survival curves were estimated according to the Kaplan-Meier method. Cox regression was used for univariate and multivariate survival analyses and results were reported as hazard ratio (HR), 95% confidence interval (CI) and p value based on statistical Wald test. A two-tailed p value of less than 0.05 indicated statistical significance. All analyses were performed using R version 3.4.1 and R-studio version 1.0.153 software.

Characteristics of the entire cohort (61 patients) are summarised in Table 1. A total of 21 patients (34%) had a DLBCL harbouring a MYC rearrangement. MYC rearrangement was observed in 11 patients (21.6%) primarily seen in the UMCG (n = 51) and 10 patients (100%) referred from affiliated hospitals (n = 10). MYC groups did not differ with regard to baseline characteristics (Table 1) except for serum LDH levels, which were higher in the MYC-positive group (p = 0.036) than in cases without MYC rearrangement.

necrosis^PET was observed in 15 patients (25%). The relationships between MYC status and necrosis^PET, necrosis^Hist and semiquantitative ¹⁸F-FDG PET parameters are summarised in Table 2. MYC⁺ cases did not differ from cases without MYC rearrangement with regard to necrosis^PET (p = 1.0) or necrosis^Hist (p = 0.52).

When the semiquantitative parameters SUV_max, SUV_max single highest, MATV and TLG were studied, no difference between MYC groups was observed. There was no relation between the presence of necrosis^PET and necrosis^Hist (p = 0.1; Supplementary Figure 1).

In 14 of 15 necrosis^PET cases, necrosis was observed in the largest lesion. In comparison, the largest individual lesion of cases without necrosis^PET had a significantly lower MATV (p = 0.0006) and SUV_max (p = 0.02), irrespective of MYC status (Supplementary Figure 2). Bulky disease was observed in 24 patients (39%). Bulky disease was significantly correlated with necrosis^PET (p = 0.005), but not with MYC status (p = 0.9) or necrosis^Hist (p = 0.8). Extranodal growth of lesions was not significantly correlated with the presence of necrosis^PET (p = 0.26).

The median follow-up was 34 months. At 5 years, OS was 67% (95% CI 54–83%), PFS was 65% (95% CI 53–81%) and DSS was 81% (95% CI 70–93%) for the entire cohort. Of the seven deaths unrelated to lymphoma, two were caused by metastatic adenocarcinoma, two were due to cardiac failure, one was due to acute on chronic renal failure and there were two cases of sudden deaths in patients in complete remission of DLBCL.

Results of the univariate Cox regression analysis (HR, 95% CI and p value) are shown in Table 3. The univariate analysis for OS identified MYC, NCCN-IPI and SUV_max single highest as associated factors. In univariate analysis for PFS, only NCCN-IPI was associated with outcome. In the univariate analysis for DSS MYC, NCCN-IPI, SUV_max single highest and necrosis^PET were associated. Both SUV_max and SUV_max single highest showed negative beta-coefficients throughout the univariate survival analysis.

For multivariate analysis, the parameters MYC, NCCN-IPI, necrosis^PET and SUV_max single highest were used due to their prognostic impact on lymphoma-related deaths in univariate analysis (Table 4). Necrosis^PET did not contribute to the prognostic model for OS and PFS. However, for DSS, necrosis^PET had a large adverse prognostic impact and proved to be independent (HR = 13.9; 95% CI 3.0–65; p = 0.001). The Kaplan-Meier analysis for DSS showed no events during the 5-year follow-up period for patients who neither had MYC rearrangements nor had necrosis^PET (n = 30) (Fig. 3).